Adding HTTPX::Response::Body#filename

This returns the filename advertised in the content-disposition header.
It reuses the same logic which existed for parsing multipart responses,
which itself was based on `rack`'s.

lib/httpx/plugins/multipart/decoder.rb

@@ -5,10 +5,6 @@ require "delegate"
 
 module HTTPX::Plugins
   module Multipart
-    using HTTPX::RegexpExtensions unless Regexp.method_defined?(:match?)
-
-    CRLF = "\r\n"
-
     class FilePart < SimpleDelegator
       attr_reader :original_filename, :content_type
 
@@ -20,32 +16,14 @@ module HTTPX::Plugins
       end
     end
 
-    TOKEN = %r{[^\s()<>,;:\\"/\[\]?=]+}.freeze
-    VALUE = /"(?:\\"|[^"])*"|#{TOKEN}/.freeze
-    CONDISP = /Content-Disposition:\s*#{TOKEN}\s*/i.freeze
-    BROKEN_QUOTED = /^#{CONDISP}.*;\s*filename="(.*?)"(?:\s*$|\s*;\s*#{TOKEN}=)/i.freeze
-    BROKEN_UNQUOTED = /^#{CONDISP}.*;\s*filename=(#{TOKEN})/i.freeze
-    MULTIPART_CONTENT_TYPE = /Content-Type: (.*)#{CRLF}/ni.freeze
-    MULTIPART_CONTENT_DISPOSITION = /Content-Disposition:.*;\s*name=(#{VALUE})/ni.freeze
-    MULTIPART_CONTENT_ID = /Content-ID:\s*([^#{CRLF}]*)/ni.freeze
-    # Updated definitions from RFC 2231
-    ATTRIBUTE_CHAR = %r{[^ \t\v\n\r)(><@,;:\\"/\[\]?='*%]}.freeze
-    ATTRIBUTE = /#{ATTRIBUTE_CHAR}+/.freeze
-    SECTION = /\*[0-9]+/.freeze
-    REGULAR_PARAMETER_NAME = /#{ATTRIBUTE}#{SECTION}?/.freeze
-    REGULAR_PARAMETER = /(#{REGULAR_PARAMETER_NAME})=(#{VALUE})/.freeze
-    EXTENDED_OTHER_NAME = /#{ATTRIBUTE}\*[1-9][0-9]*\*/.freeze
-    EXTENDED_OTHER_VALUE = /%[0-9a-fA-F]{2}|#{ATTRIBUTE_CHAR}/.freeze
-    EXTENDED_OTHER_PARAMETER = /(#{EXTENDED_OTHER_NAME})=(#{EXTENDED_OTHER_VALUE}*)/.freeze
-    EXTENDED_INITIAL_NAME = /#{ATTRIBUTE}(?:\*0)?\*/.freeze
-    EXTENDED_INITIAL_VALUE = /[a-zA-Z0-9-]*'[a-zA-Z0-9-]*'#{EXTENDED_OTHER_VALUE}*/.freeze
-    EXTENDED_INITIAL_PARAMETER = /(#{EXTENDED_INITIAL_NAME})=(#{EXTENDED_INITIAL_VALUE})/.freeze
-    EXTENDED_PARAMETER = /#{EXTENDED_INITIAL_PARAMETER}|#{EXTENDED_OTHER_PARAMETER}/.freeze
-    DISPPARM = /;\s*(?:#{REGULAR_PARAMETER}|#{EXTENDED_PARAMETER})\s*/.freeze
-    RFC2183 = /^#{CONDISP}(#{DISPPARM})+$/i.freeze
-
     class Decoder
+      include HTTPX::Utils
+
+      CRLF = "\r\n"
       BOUNDARY_RE = /;\s*boundary=([^;]+)/i.freeze
+      MULTIPART_CONTENT_TYPE = /Content-Type: (.*)#{CRLF}/ni.freeze
+      MULTIPART_CONTENT_DISPOSITION = /Content-Disposition:.*;\s*name=(#{VALUE})/ni.freeze
+      MULTIPART_CONTENT_ID = /Content-ID:\s*([^#{CRLF}]*)/ni.freeze
       WINDOW_SIZE = 2 << 14
 
       def initialize(response)
@@ -102,7 +80,7 @@ module HTTPX::Plugins
             name = head[MULTIPART_CONTENT_ID, 1]
           end
 
-          filename = get_filename(head)
+          filename = HTTPX::Utils.get_filename(head)
 
           name = filename || +"#{content_type || "text/plain"}[]" if name.nil? || name.empty?
 
@@ -154,34 +132,6 @@ module HTTPX::Plugins
           raise Error, "parsing should have been over by now"
         end until @buffer.empty?
       end
-
-      def get_filename(head)
-        filename = nil
-        case head
-        when RFC2183
-          params = Hash[*head.scan(DISPPARM).flat_map(&:compact)]
-
-          if (filename = params["filename"])
-            filename = Regexp.last_match(1) if filename =~ /^"(.*)"$/
-          elsif (filename = params["filename*"])
-            encoding, _, filename = filename.split("'", 3)
-          end
-        when BROKEN_QUOTED, BROKEN_UNQUOTED
-          filename = Regexp.last_match(1)
-        end
-
-        return unless filename
-
-        filename = URI::DEFAULT_PARSER.unescape(filename) if filename.scan(/%.?.?/).all? { |s| /%[0-9a-fA-F]{2}/.match?(s) }
-
-        filename.scrub!
-
-        filename = filename.gsub(/\\(.)/, '\1') unless /\\[^\\"]/.match?(filename)
-
-        filename.force_encoding ::Encoding.find(encoding) if encoding
-
-        filename
-      end
     end
   end
 end

lib/httpx/response.rb

@@ -181,6 +181,12 @@ module HTTPX
         end
       end
 
+      def filename
+        return unless @headers.key?("content-disposition")
+
+        Utils.get_filename(@headers["content-disposition"])
+      end
+
       def to_s
         case @buffer
         when StringIO

lib/httpx/utils.rb

@@ -3,6 +3,12 @@
 module HTTPX
   module Utils
     using URIExtensions
+    using HTTPX::RegexpExtensions unless Regexp.method_defined?(:match?)
+
+    TOKEN = %r{[^\s()<>,;:\\"/\[\]?=]+}.freeze
+    VALUE = /"(?:\\"|[^"])*"|#{TOKEN}/.freeze
+    FILENAME_REGEX = /\s*filename=(#{VALUE})/.freeze
+    FILENAME_EXTENSION_REGEX = /\s*filename\*=(#{VALUE})/.freeze
 
     module_function
 
@@ -25,6 +31,30 @@ module HTTPX
       time - Time.now
     end
 
+    def get_filename(header, _prefix_regex = nil)
+      filename = nil
+      case header
+      when FILENAME_REGEX
+        filename = Regexp.last_match(1)
+        filename = Regexp.last_match(1) if filename =~ /^"(.*)"$/
+      when FILENAME_EXTENSION_REGEX
+        filename = Regexp.last_match(1)
+        encoding, _, filename = filename.split("'", 3)
+      end
+
+      return unless filename
+
+      filename = URI::DEFAULT_PARSER.unescape(filename) if filename.scan(/%.?.?/).all? { |s| /%[0-9a-fA-F]{2}/.match?(s) }
+
+      filename.scrub!
+
+      filename = filename.gsub(/\\(.)/, '\1') unless /\\[^\\"]/.match?(filename)
+
+      filename.force_encoding ::Encoding.find(encoding) if encoding
+
+      filename
+    end
+
     if RUBY_VERSION < "2.3"
 
       def to_uri(uri)

sig/plugins/multipart.rbs

@@ -68,8 +68,6 @@ module HTTPX
         def initialize: (Response response) -> void
 
         def parse: () -> void
-
-        def get_filename: (String head) -> String?
       end
 
       class FilePart #  < SimpleDelegator

sig/response.rbs

@@ -64,6 +64,8 @@ module HTTPX
       def each: () { (String) -> void } -> void
               | () -> Enumerable[String]
 
+      def filename: () -> String?
+
       def bytesize: () -> (Integer | Float)
       def empty?: () -> bool
       def copy_to: (String | File | _Writer destination) -> void

sig/utils.rbs

@@ -9,5 +9,7 @@ module HTTPX
     def self?.elapsed_time: (Integer | Float monotonic_time) -> Float
 
     def self?.to_uri: (generic_uri uri) -> URI::Generic
+
+    def self?.get_filename: (String header) -> String?
   end
 end

test/response_test.rb

@@ -138,6 +138,33 @@ class ResponseTest < Minitest::Test
     assert body.buffer.is_a?(Tempfile), "body should buffer to file after going over threshold"
   end
 
+  def test_response_body_filename
+    body = Response::Body.new(Response.new(request, 200, "2.0", {}), Options.new)
+    assert body.filename.nil?
+    body = Response::Body.new(Response.new(request, 200, "2.0", { "content-disposition" => "attachment;filename=test.csv" }), Options.new)
+    assert body.filename == "test.csv"
+    body = Response::Body.new(Response.new(request, 200, "2.0", { "content-disposition" => "attachment;filename=\"test.csv\"" }),
+                              Options.new)
+    assert body.filename == "test.csv"
+    body = Response::Body.new(Response.new(request, 200, "2.0", {
+                                             "content-disposition" => "inline; filename=ER886357.pdf; " \
+                                                                      "creation-date=9/17/2012 1:51:37 PM; " \
+                                                                      "modification-date=9/17/2012 1:51:37 PM; size=3718678",
+                                           }),
+                              Options.new)
+    assert body.filename == "ER886357.pdf"
+
+    body = Response::Body.new(Response.new(request, 200, "2.0", { "content-disposition" => "attachment; filename*=UTF-8''bar" }),
+                              Options.new)
+    assert body.filename == "bar"
+    body = Response::Body.new(Response.new(request, 200, "2.0", {
+                                             "content-disposition" => "inline; filename*=UTF-8''%c2%a3%20and%20%e2%82%ac%20rates.pdf",
+                                           }),
+                              Options.new)
+
+    assert body.filename == "£ and € rates.pdf"
+  end
+
   def test_response_decoders
     json_response = Response.new(request, 200, "2.0", { "content-type" => "application/json" })
     json_response << %({"a": "b"})